As I scrolled through my social media feeds over the Christmas break Ireadthroughseveral “what do AI tools like ChatGPT mean for the future of Higher Education” pieces—I’m sure you saw them as well. To be honest, I’m reluctant to add to that discourse, because most of my thoughts on the issue have already been made.

I was at a BBQ dinner with a colleague last weekend and our chat turned to “how’s your class shaping up for the coming semester?” My colleague was aware of ChatGPT and isn’t worried about it destroying the fabric of higher education or anything like that, but they were a bit unsure about if and how they could make use of it in their class. So here are a few things to think about if you’re in that position. I’m going to use ChatGPT as an example (because it’s the hot one right now), but similar questions apply for any AI content generation tool, whether for generating text images, music, voiceover or video.

First, if your class includes student deliverables which involve submitting written work (e.g. essays, but also lab reports, process blogs, compulsory forum posts and more) then some of your students are going to use ChatGPT to write their submissions. It’s inevitable, and although the NY Schools district is trying to bury its head in the sand and ban ChatGPT, this is (in my opinion) both wrongheaded and too difficult to enforce in practice.

So, as you look over your course outline and assessment schedule you need to ask yourself whether you care if your students use ChatGPT. It’s a question that goes to the heart of why we set deliverables at all, which includes reasons like legal compliance but for most educators is also about deeply held ideals about education as the pursuit of knowledge and truth, as well as what it means to give a good (or bad) grade.

Maybe you don’t care, or at least you don’t care enough to re-jig your whole class to “defend” against students using ChatGPT to submit work that they don’t understand. I think that’s an ok position to take for now, although I’d encourage you to watch closely as other educators try things (and no doubt fail in interesting and unforseen ways). We will learn from each others’ experiences in how best to incorporate these tools in our classrooms. But you should at least think hard about “why do I set that particular essays as the final assessment item in this course, and what would it mean if a student used ChatGPT to write it?”

Secondly, there are already some good ideas floating around about how to use ChatGPT productively in your teaching. This isn’t just if you’re designing a new course on “Applied AI Language Models”, there are ways to incorporate it into an existing course as well. The Centre for Learning and Teaching at Washington University has already got a helpful ChatGPT and AI Composition Tools page up on their website. There are a bunch of folks experimenting with ChatGPT-designed syllabi.

I think that people like Yoav Goldberg are on the right track when they say

“actually now that i think of it more, i think a “write an essay with the help of chatGPT and discuss the process and the resulting prose” can be a super-effective assignment also in a humanities-centric, critical-ai program.”

If you’re wondering how you might incorporate ChatGPT into your class this coming semester, look through your course outline for every time you ask your students to produce a written artefact in response to a question or prompt that you give them. Can you change it up so that the student (or students) asks that question of ChatGPT and then has to critically reflect on the output they recieve from the AI model? This process could be scaffolded by first doing it in a facilitated class discussion setting, but then later as an assessment task (and perhaps the students need to create their own rubric or criteria on which the output of the AI model should be evaluated). There are heaps of variations on this idea, but the general idea is this: wherever in your current class you ask your students to write something, that’s an opportunity to get the student to co-write (or evaluate) that same something with ChatGPT. On a superficial level this sort of “AI tool use” is the sort of process that knowledge workers will increasingly be incorporating into our daily workflows. At a deeper level it’s an opportunity to reflect on where these tools work well, where they’re useless, and where they might be actively harmful.

One final point which I made to my colleague in the context of incorporating ChatGPT into this semesters’s class is about availability. During the current initial research preview ChatGPT is free to use. But OpenAI (creators of ChatGPT) haven’t released any information about how long this research preview window will last, or how much it’ll cost when that window ends.

ChatGPT running costs are estimated at $3m USD per month, and OpenAI are a private company, so they can’t be counted on to keep it running free forever as a public good. If it’s eventually priced similarly to GPT3 (the AI language model on which it’s based) then it’ll be fairly cheap—probably only a few cents to write an article as long as this one—but you’re still left with an equity issue unless you’re willing to pay for a subscription for all your students. Even if your institution is willing, these tools don’t yet have an (easy) way to sign up a whole class, ensuring that student’s activity stays under a given budget, and send one easy bill. I’m sure that will come in time, and AI companies may well be willing to offer partnerships and scholarships with educational institutions doing this sort of thing, but be prepared for a small panic if ChatGPT’s free trial gets turned off 24 hours before an assessment deadline.

I’m not trying to play the techno-optimist and downplay the potential downsides of ChatGPT in higher education. These tools may well blow up society, and higher education is (clearly!) not going to be unaffected if that happens. My main point is that if you’re careful you can start using them in the classroom today—your students will be anyway, and this way you get to go on that journey with them. There will be some challenges, and you might have to ask yourself some deep questions about why your class is designed the way it is. But it is a pretty exciting chance to be part of a community of educators figuring it out together, and whatever conclusions we come to I do strongly believe that we need to learn how to live in a world of AI-generated bullshit, so why wouldn’t we set our students up to start that learning journey with us while they’re in our classrooms?